A Comparative Analysis of Supervised Multi-label Text Classification Methods

Multi-label classification methods are getting more popular now a days because of their increasing demand in various application domains such as text classification , image classification , functional genomics , music categorization , emotion recognition etc. Multi-label classification methods are falling under two broader categories of problem transformation methods and algorithm adaptation methods. From machine learning perspective both of these types are working under the roof of supervised classification methods wherein the labels are already provided in the training data set. An attempt is made through this paper to present the state of the art supervised text classification techniques and there comparison. The paper also discusses the important results reported so far in text classification domain and also tried to highlight the beneficial directions of the research till date. The experiments are conducted on standard bench mark datasets such as Enron, Bibtex and Slashdot. Moreover, the paper also contains a comprehensive bibliography of selected papers appeared in reputed journals and conference proceedings as an aid for the researchers working in the field of multi-label classification domain.